As part of our ongoing study of the evolved giants in the galactic halo, we consider the sample of 1286 long-period variables (LPVs) in the southern hemisphere provided by the Catalina Sky Survey experiment. These LPVs have periods P larger than 80 days and amplitudes over 0.2 mag. First, by using the Hamburg-ESO spectral survey, we aim to determine the spectral type as either M-type or C-type for objects located in the imprint of this survey, $|b| > 30$ degrees. Of 135 LPVs obeying this selection, we classified 93, and found only 2 new carbon stars. Secondly, we consider faint LPVs. We discovered that many lie at about 1 arcmin from a bright Mira catalogued in the General Catalog of Variable Stars, with identical period. We study these suspicious cases in detail, and conclude that, for as many as 56 faint Catalina LPVs, their variability is due to contamination by light from the bright, neighbouring GCVS Mira: an instrumental artefact. We conclude that when dealing with distant, faint Miras in the Catalina catalog, researchers should pay attention to the polluting effects of neighbouring bright and variable objects.